4_cfg.json

{"seed": 49, "batch_size": 4096, "buffer_mult": 128, "lr": 5e-05, "num_tokens": 191931392, "l1_coeff": 2, "beta1": 0.9, "beta2": 0.999, "d_in": 1536, "dict_size": 16384, "seq_len": 1024, "enc_dtype": "fp32", "model_name": "qwen2.5-math-1.5b", "site": "resid_pre", "device": "cuda", "model_batch_size": 4, "log_every": 20, "save_every": 10000, "dec_init_norm": 0.08, "hook_point": "blocks.14.hook_resid_pre", "wandb_project": "R1-crosscoder", "wandb_entity": "Neelectric", "run_name": "qwen2.5-math-1.5b_crosscoder"}